To illustrate several of the above points made concerning development of a high throughput enzyme assay for natural product screening, the development of an enzyme assay for farnesyltransferase will be presented.
Farnesyltransferase is an enzyme that catalyzes the addition of a C-15 farnesyl
isoprenoid moiety onto specific protein substrates possessing a C-terminal amino acid CAAX box (Casey et al., 1989). Ras is one physiological substrate for farnesyltransferase and this modification has been shown to be essential for Ras associated cellular transformation [Cox et al., 1992; Schafer et al., 1992]. Several forms of human cancer that are resistant to conventional anticancer drugs, such as colon and pancreatic cancers, contain high levels of mutated Ras protein [Lowry et al., 1993]. Due to these observations, it has been postulated that inhibitors of Ras farnesylation may be valuable anticancer agents. Several different peptide-like and natural product inhibitors of farnesyltransferase have been identified in screening programs and some of these have been reported to inhibit transformation of normal cells by mutant Ras proteins [Gibbs et al., 1994].
Farnesyltransferase is a zinc containing cytosolic heterodimeric protein containing - and -subunits which requires both zinc and magnesium ions for activity [Chen et al., 1991; Reiss et al., 1990]. The -subunit binds farnesylpyrophosphate and is shared with the related isoprenylation enzyme, geranylgeranyltransferase. The -subunit binds to substrate. It is structurally distinct from the -subunit of geranylgeranyltransferase and binds to the substrate that becomes isoprenylated. The farnesyltransferase assay is thus a two-substrate enzymatic reaction requiring both farnesylpyrophosphate and protein substrate containing the N-terminal CAAX box. Detailed kinetic studies have indicated that there is a random order of substrate addition to the enzyme (Pompliano et al., 1992). In configuring the assay for drug discovery, a decision needed to be made as to what type of inhibitor the screen should be biased to identify. Competitive inhibitors of farnesylpyrophopshate binding would not be desired because of a potential cross- reactjvity of these inhibitors with geranylgeranyl-transferase, resulting in a lack of selectivity and possibly severe adverse effects. A screen designed to identify competitive inhibitors of substrate binding to the -subunit would be more desirable because the potential for selectivity for only farnsyltransferase would be much higher.
The first consideration for developing an assay was to identify an appropriate source of the farnesyltransferase enzyme. Rat brain was chosen because no suitable human sources
could be identified. However, human placenta has been recently suggested to be a good source of human enzyme. The enzyme was partially purified (50 fold) by subjecting 100,000×g supernat ants to ammonium sulfate precipitation and MonoQ and Sephacel S- 200 column chromatography. A normal preparation using 50 rat brains provided enough material for approximately 100 assay plates. These preparations were checked for geranylgeranyltransferase activity and none was detected. Studies of enzyme stability indicated that the partially purified enzyme was stable with storage at −80°C for a period of one month. Subsequent to developing this assay, a patent was awarded to the University of Texas Southwestern Medical Center covering the isolation of farnesyltransferase. This technology can be licensed for use in drug discovery.
Several different assays for farnesyltransferase could be considered for developing a screening assay. The most attractive from a high throughput standpoint was a farnesyltransferase scintillation proximity assay kit commercially available from Amersham. The major attraction of this assay was that a separation step normally required to resolve substrates and product in typical radiometric assay methods is not required in the SPA assay. Figure 9 shows a diagramatic representation of the principle and the steps associated with the SPA assay.
Figure 9: Principle of the Amersham SPA farnesyltransferase assay.
The principle of the SPA assay is that the [3H] farnesyl moiety is incorporated into a biotinylated synthetic peptide corresponding to the C-terminal tail of lamin, a physiological substrate for farnesyltransferase. After incorporation, streptavidin coupled to SPA beads is added. By virtue of the high affinity interaction of streptavidin and biotin, the [3H]-farnesyl moiety incorporated into the lamin peptide is brought into close proximity with the scintillant containing SPA bead. Unincorporated [3H]- farnesyl moiety not in close proximity with the SPA beads, does not contribute to the signal. The actual assay steps involved in the SPA assay are as follows:
1. 50µl of assay buffer, 15µl of 0.5µM biotinyl-lamin peptide, and 15µl of 60µM [3H]- farnesylpyrophos-phate are added to microtiter plates. Negative controls also contain 2 µM CVLS peptide.
3. Test compounds are added in 10 µl aliquots.
5. The enzyme reaction is initiated by the addition of 10 µl of enzyme ( 1µg). 7. Plates are incubated for precisely 60 minutes at 37°C.
9. 150 µl of stop buffer containing streptavidin-coated SPA beads is added to each well and the plates are incubated for 30 minutes at room temperature.
11. Plates are then read in a microtiter plate scintillation counter.
This assay is well suited for high volume screening in that only two addition steps are required before endpoint reading. Since the assay is in a micotiter plate format, it is amenable to robotic manipulation and can be completely automated.
Although the assay kit provided by Amersham contains an assay protocol, it is necessary to optimize the assay for screening. The first step in the assay development and validation is to evaluate the relation of enzyme concentration to the assay signal-to-noise ratio. This experiment was performed in the presence of saturating concentrations of [3H]-farnesylpyrophosphate and a concentration of biotinyl-lamin which was slightly higher than the Km value for enzyme. Experimental data indicated that the activity was linear up to 2 µg of added protein. At 1 µg, the signal-to-noise was 100:1 and acceptable for the screening assay (Figure 10).
Figure 10: Variable addition of farnesytransferase in the SPA assay. Final
concentrations of 9 mM [3H]-FPP, 100 nM biotinyl-lamin peptide, and indicated concentrations of partially purified enzyme were present in the assay and incubations were performed for 60 minutes at room temperature. Data points represent the mean+/− SEM of triplicate determinations at each enzyme concentration. Data were fit to a hyperbolic function using nonlinear least squares curve fitting program, Prism.
The next experiment was designed to determine the Km value for [3H]- farnesylpyrophosphate and was performed in the presence of a saturating concentration of biotinyl-lamin (0.075 µM) and 1 µg of the enzyme. Results of this experiment indicated that [3H]-farnesylpyrophosphate had a Km value of approximately 2.3 µM (Figure 11). A decision was made to include 10 µM final concentration of [3H]-
farnesylpyrophosphate to ensure that the enzyme was fully saturated with respect to this substrate and more biased in detecting inhibitors of biotinyl-lamin interaction with farnesyltransferase.
Figure 11: Variable addition of [3H]-farnesylpyrophosphate in the SPA farnesyltransferase assay. Final concentrations of Img of enzyme, 100 nM biotinyl-lamin, and indicated concentrations of [3H]-
farnesylpyrophosphate were present in the assay; incubations were performed for 60 minutes at room temperature. Data points represent the mean+/− SEM of triplicate determinations made at each [3H]- FPP concentration. Data were fit to a hyperbolic function using the nonlinear least squares curve fitting program, Prism.
The Km value for biotinyl-lamin was next determined in the presence of a saturating concentration of [3H]- farnesylpyrophosphate and 1 µg of enzyme (Figure 12). A Km value of 19 nM was determined. Due to a relatively low signal-to-noise at 20 nM, a decision was made to include 70 nM biotinyl-lamin in the assay to achieve a high signal- to-noise ratio in the range of 100:1.
A time course evaluation was next performed in the presence of 10 [3H]-
farnesylpyrophosphate, 70 nM biotinyl-lamin, and 1 µg of enzyme/well (Figure 13). Results of this evaluation indicated that the % [3H]-farnesylpyrophosphate incorporated increased linearly between 10 and 60 minutes at room temperature. An incubation period of 60 minutes at room temperature was chosen for subsequent assays. At this time period, incubations are under initial rate conditions.
Figure 12: Variable addition of biotinyl-lamin in the SPA farnesyltransferase
assay. Final concentrations of Img of enzyme, 10 mM [3H]-FPP, and indicated concentrations of biotinyl-lamin were present; incubations were performed for 60 minutes at room temperature. Data points represent the mean+/−SEM of triplicate determinations made at each indicated biotinyl-lamin concentration. Data were fit to a hyperbolic function using the nonlinear least squares curve fitting program, Prism.
During development of this assay, no inhibitors were commercially available for evaluation in the assay. A tetrapeptide CVSL, which corresponds to the CAAX box of p21-Harvey Ras, was used to inhibit farnesylation of biotinyl-lamin. Since CVLS is a substrate for farnesyltransferase, it would compete for farnesylation with biotinyl-lamin. This peptide inhibited incorporation of [3H]-farnesylpyrophosphate into biotinyl-lamin with an IC50 value of 796 nM (Figure 14).
The influence of DMSO on farnesyltransferase activity was evaluated next to determine the assay tolerance for DMSO. Activity was unaffected by DMSO concentrations upto 10% and this value of final DMSO concentration was used in the screening assay (data not shown). Several different types of ethyl acetate extracts were evaluated in the assay to establish the final concentration of extract that would result in a hit rate of approximately 0.5%. A final concentration of 0.1X (of the original 10X concentrated extract) for 200 different terrestrial and marine actinomycete extracts and 200 different terrestrial and marine fungal extracts was found to generate the desired assay hit rate. Organic methylene chloride extracts, aqueous ethanol extracts, and
aqueous ethanol extracts subjected to poly phenol removal by chromatography prepared from for 100 different plants were also tested in the assay. A final assay concentration of 40 mg/ml for all three different plant extracts gave a hit rate of approximately 0.5%.
Figure 13: Farnesyltransferase assay time course. The assay contained final
concentrations of 1 mg enzyme, 10 mM [3H]-FPP, and 70 nM biotinyl-lamin and incubations were performed for indicated periods at room temperature. Triplicate determinations were made at each time point, the mean was calculated, and the percentage of added [3H]-FPP that was incorporated into substrate is indicated. The line drawn represents linear regression of the data using the nonlinear least squares curve fitting program, Prism.